One of the assumptions of the Market Model is that firm specific events are uncorrelated across assets.
If the above is true then the Cov will always be zero - using the covariance formula for Market Model we have:
Cov for asset i and j = Beta i * Beta j* Variance for market
= Beta i * Beta j * (Correlation for i and j * std dev i * std dev j) / Beta i
So, if we subtitute zero for 'Correlation for i and j' above then the 'Cov for asset i and j' is zero.
Am I missing something?作者: jarobi04 时间: 2011-7-11 20:06
Well I am no expert on this, I plan on studying it towards the end just cause it so big and has little weight.
BUT
I dont see how you concluded that cov will be 0...
Firm specific events are uncorrolated, but both assets are still corrolated to market events via their Beta, and that can be used to get the cov of the two assets....作者: economicz 时间: 2011-7-11 20:06
Let's work through an example:
If a firm specific event impacts asset i, the components which make up the event should be uncorrelated across components of other assets - however asset i as a whole CAN be correlated with another asset through Beta.
So, according to the market model - in order for the covariance to be correct the components should be uncorrelated - if they are then the covariance is incorrect.
If THAT is our understanding - then how do we test for the incorrectness of market model covariance in the cases where the components are in fact correlated, i.e. do we still use the market model covariance?作者: wxs1986 时间: 2011-7-11 20:06
good explanation gulf thanks. Just think "market" model.....beta is basically the market....so do beta*beta*variance of mkt ?作者: kkn006 时间: 2011-7-11 20:06
^ the point is under the assumptions of the model the covar will be "beta*beta*variance of mkt"
now me and you can sit here and talk about it all day claiming to understand it, but actually the only way to claim such (or so i belive) is to sit down and derive that formula, or at least read the derivation and see that it makes sense...
a lot of people miss the point that finance is mostly math...
anyway, i doubt any of us has the time given that we are so close to the exam to actually do this, and most of us (possibly myself) dont have the ability to do it anyway..
so for exam purposes, just learn it as a formula that you use to estimate covar and life goes on
